summaryrefslogtreecommitdiff
path: root/tools/audio/extraction/tuning.py
blob: 065cd714ffd1259776990ffb2405a36a963cba4c (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
#!/usr/bin/env python3
# SPDX-FileCopyrightText: © 2024 ZeldaRET
# SPDX-License-Identifier: CC0-1.0
#
#   Estimate (samplerate, basenote) from tuning
#
#   tuning = samplerate * 2 ** basenote
#

from typing import List, Tuple

from .util import f32, u32_to_f32, f32_to_u32

# Mirrors gPitchFrequencies in audio driver source.
# Indexed by z64 note numbers, g_pitch_frequencies[C4] = 1.0 (0x3F800000)
# Converted to their IEEE-754 binary representation to avoid any string -> float parser trouble as we need exact values.
g_pitch_frequencies = (
    0x3DD744F6, 0x3DE411C3, 0x3DF1A198, 0x3E000000, 0x3E079C84, 0x3E0FACE6, 0x3E1837F8, 0x3E21450F,
    0x3E2ADC0A, 0x3E350508, 0x3E3FC86D, 0x3E4B2FEC, 0x3E5744F6, 0x3E641206, 0x3E71A1DC, 0x3E800000,
    0x3E879C84, 0x3E8FACE6, 0x3E9837F8, 0x3EA1450F, 0x3EAADC0A, 0x3EB504E6, 0x3EBFC88E, 0x3ECB2FEC,
    0x3ED744F6, 0x3EE411E4, 0x3EF1A1BA, 0x3F000000, 0x3F079C84, 0x3F0FACD6, 0x3F1837F8, 0x3F214520,
    0x3F2ADC0A, 0x3F3504F7, 0x3F3FC88E, 0x3F4B2FFD, 0x3F574507, 0x3F6411F5, 0x3F71A1CB, 0x3F800000,
    0x3F879C7C, 0x3F8FACD6, 0x3F9837EF, 0x3FA14517, 0x3FAADC0A, 0x3FB504F7, 0x3FBFC886, 0x3FCB2FF5,
    0x3FD744FE, 0x3FE411F5, 0x3FF1A1C2, 0x40000000, 0x40079C7C, 0x400FACD6, 0x401837EF, 0x40214517,
    0x402ADC0A, 0x403504F7, 0x403FC88A, 0x404B2FF9, 0x405744FE, 0x406411F5, 0x4071A1C2, 0x40800000,
    0x40879C7E, 0x408FACD8, 0x409837F1, 0x40A14519, 0x40AADC0A, 0x40B504F5, 0x40BFC888, 0x40CB2FF9,
    0x40D74500, 0x40E411F5, 0x40F1A1C2, 0x41000000, 0x41079C7D, 0x410FACD7, 0x411837F1, 0x41214519,
    0x412ADC0A, 0x413504F5, 0x413FC889, 0x414B2FF8, 0x41574500, 0x416411F4, 0x4171A1C3, 0x41800000,
    0x41879C7D, 0x418FACD7, 0x419837F1, 0x41A14519, 0x41AADC0A, 0x41B504F5, 0x41BFC889, 0x41CB2FF8,
    0x41D74500, 0x41E411F4, 0x41F1A1C3, 0x42000000, 0x42079C7D, 0x420FACD7, 0x421837F1, 0x42214519,
    0x422ADC0A, 0x423504F5, 0x423FC889, 0x424B2FF8, 0x42574500, 0x426411F4, 0x4271A1C3, 0x42800000,
    0x42879C7D, 0x428FACD7, 0x429837F1, 0x42A14519, 0x42AADC0A, 0x3D6411C3, 0x3D71A198, 0x3D800000,
    0x3D879C41, 0x3D8FACE6, 0x3D9837B5, 0x3DA1450F, 0x3DAADBC6, 0x3DB504C5, 0x3DBFC86D, 0x3DCB302F,
)

# Names for pitch values indexed by z64 note numbers, pitch_names[39] = C4
pitch_names = (
                                                                          "A0", "BF0", "B0",
    "C1",  "DF1",  "D1",  "EF1",  "E1",  "F1",  "GF1",    "G1",    "AF1", "A1", "BF1", "B1",
    "C2",  "DF2",  "D2",  "EF2",  "E2",  "F2",  "GF2",    "G2",    "AF2", "A2", "BF2", "B2",
    "C3",  "DF3",  "D3",  "EF3",  "E3",  "F3",  "GF3",    "G3",    "AF3", "A3", "BF3", "B3",
    "C4",  "DF4",  "D4",  "EF4",  "E4",  "F4",  "GF4",    "G4",    "AF4", "A4", "BF4", "B4",
    "C5",  "DF5",  "D5",  "EF5",  "E5",  "F5",  "GF5",    "G5",    "AF5", "A5", "BF5", "B5",
    "C6",  "DF6",  "D6",  "EF6",  "E6",  "F6",  "GF6",    "G6",    "AF6", "A6", "BF6", "B6",
    "C7",  "DF7",  "D7",  "EF7",  "E7",  "F7",  "GF7",    "G7",    "AF7", "A7", "BF7", "B7",
    "C8",  "DF8",  "D8",  "EF8",  "E8",  "F8",  "GF8",    "G8",    "AF8", "A8", "BF8", "B8",
    "C9",  "DF9",  "D9",  "EF9",  "E9",  "F9",  "GF9",    "G9",    "AF9", "A9", "BF9", "B9",
    "C10", "DF10", "D10", "EF10", "E10", "F10",
                                                                                "BFNEG1", "BNEG1",
    "C0",  "DF0",  "D0",  "EF0",  "E0",  "F0",  "GF0",    "G0",    "AF0",
)

# Floats that are encountered in extraction but cannot be resolved to a match.
BAD_FLOATS = [0x3E7319E3]

def note_z64_to_midi(note : int) -> int:
    """
    Convert a z64 note number to MIDI note number.

    Middle C is 39 in z64, while it is 60 in MIDI.
    We want MIDI note numbers to store in the extracted sample files (aiff or wav)
    """
    return (21 + note) % 128

def recalc_tuning(rate : int, note : int) -> float:
    # The tuning formula t(r,n) for n a midi note number is
    #   t = (r / 32000) * 2^{60 - n}
    # We use a lookup table for the power of 2 calculation for z a z64 note number
    #   t = (r / 32000) * frequencies[78 - z]
    # The offset by 78 comes from 2*(60 - 21) where 21 relates the z64 and midi note numbers
    #   n = 21 + z
    return f32(f32(rate / 32000.0) * u32_to_f32(g_pitch_frequencies[(78 - note) % 128]))

def rate_from_tuning(tuning : float) -> Tuple[Tuple[str,int]]:
    """
    Decompose a tuning value into a pair (samplerate, basenote) that round-trips when ran through `recalc_tuning`
    """
    matches : List[Tuple[str,int]] = []
    diffs : List[Tuple[int, Tuple[str,int]]] = []

    tuning_bits : int = f32_to_u32(tuning)

    def test_value(note_val : int, nominal_rate : int, freq : float):
        if nominal_rate > 48000:
            # reject samplerate if too high
            return

        # recalc tuning and compare to original

        tuning2 : float = f32(f32(nominal_rate / 32000.0) * freq)

        diff : int = abs(f32_to_u32(tuning2) - tuning_bits)

        if diff == 0:
            matches.append((note_val, nominal_rate))
        else:
            diffs.append((diff, (note_val, nominal_rate)))

    # search gPitchFrequencies LUT one by one. We don't exit as soon as a match is found as in general this procedure
    # only recovers the correct (rate,note) pair up to multiples of 2, to get the final value we want to select the
    # "best" of these pairs by an essentially arbitrary ranking (cf `rank_rates_notes`)
    for i,freq_bits in enumerate(g_pitch_frequencies):
        # Reflect the note value as in recalc_tuning
        note_val = (78 - i) % 128
        freq : float = u32_to_f32(freq_bits)

        # compute the "nominal" samplerate for a given basenote by R = 32000 * (t / f)
        nominal_rate : int = int(f32(tuning / freq) * 32000.0)

        # test nominal value and +/-1
        test_value(note_val, nominal_rate,     freq)
        test_value(note_val, nominal_rate + 1, freq)
        test_value(note_val, nominal_rate - 1, freq)

    if len(matches) != 0:
        return tuple(matches)

    # no matches found... check if we expected this, otherwise flag it for special handling
    assert tuning_bits in BAD_FLOATS , f"0x{tuning_bits:08X}"

    # just take the closest match and hack it in the soundfont compiler
    hack_rate = sorted(diffs, key=lambda e : e[0])[0]
    return (hack_rate[1],)

def rank_rates_notes(layouts):

    def rank_rate_note(rate, notes):
        """
        Arbitrarily rank the input samplerate + note numbers, based on what is most likely.
        """
        rank = 0

        notes_named = [pitch_names[note] for note in notes]

        if 'C4' in notes_named and rate > 10000:
            rank += 10000
        elif 'C2' in notes_named and rate > 10000:
            rank += 9500
        elif 'D3' in notes_named and rate > 10000:
            rank += 8500
        elif 'D4' in notes_named and rate > 10000:
            rank += 8000
        elif 'C3' in notes_named and rate > 10000:
            rank += 4000
        elif 'C5' in notes_named and rate > 10000:
            rank += 4000
        elif 'D0' in notes_named:
            rank += 3500
        elif 'A9' in notes_named:
            rank += 3000
        elif 'A3' in notes_named:
            rank += 2750
        elif 'C6' in notes_named:
            rank += 2750
        elif 'C8' in notes_named:
            rank += 2500
        elif 'C1' in notes_named:
            rank += 2500
        elif 'A4' in notes_named:
            rank += 2500
        elif 'A0' in notes_named:
            rank += 2250
        elif 'B0' in notes_named:
            rank += 2250
        elif 'AF9' in notes_named:
            rank += 2000
        elif 'AF0' in notes_named:
            rank += 2000
        elif 'G3' in notes_named:
            rank += 2000
        elif 'GF4' in notes_named:
            rank += 100
        elif 'F9' in notes_named:
            rank += 50
        elif 'F3' in notes_named:
            rank += 25
        elif 'C0' in notes_named:
            rank += 50
        elif 'BF2' in notes_named:
            rank += 30
        elif 'B3' in notes_named:
            rank += 25
        elif 'BF1' in notes_named:
            rank += 25
        elif 'E2' in notes_named:
            rank += 20
        elif 'F6' in notes_named:
            rank += 15
        elif 'GF2' in notes_named:
            rank += 10
        elif 'BF3' in notes_named:
            rank += 1
        elif 'AF2' in notes_named:
            rank += 1

        rank += {
            32000 : 200,
            16000 : 100,
            24000 : 50,
            22050 : 30,
            20000 : 28,
            44100 : 25,
            12000 : 15,
            8000  : 10,
            15950 : 5,
            20050 : 5,
            31800 : 5,
        }.get(rate, 0)

        return rank

    # Input should not be empty
    assert len(layouts) != 0

    if len(layouts) == 1:
        # No ranking needed, there is only one possible option
        return layouts[0]

    # Ranking is needed, rank each layout
    ranked = list(sorted(layouts, key=lambda L : rank_rate_note(*L), reverse=True))

    # Ensure the ranking produced a unique best option
    assert rank_rate_note(*ranked[0]) != rank_rate_note(*ranked[1]) , [(rate,tuple(pitch_names[note] for note in notes)) for rate,notes in ranked]

    # Output best
    return ranked[0]

if __name__ == '__main__':
    import argparse

    parser = argparse.ArgumentParser(description="Given either a (rate,note) or a tuning, compute all matching rates/notes.")
    parser.add_argument("-t", dest="tuning", required=False, default=None, help="Tuning value (float or hex)")
    parser.add_argument("-r", dest="rate", required=False, default=None, type=int, help="Sample rate (integer)")
    parser.add_argument("-n", dest="note", required=False, default=None, type=str, help="Base note (note name)")
    parser.add_argument("--show-result", required=False, default=False, action="store_true", help="Show recalculated tuning value")
    args = parser.parse_args()

    if args.tuning is not None:
        # Take input tuning
        tuning : float = u32_to_f32(int(args.tuning,16)) if args.tuning.startswith("0x") else float(args.tuning)
    elif args.rate is not None and args.note is not None:
        # Calculate target tuning from input rate and note
        tuning : float = recalc_tuning(args.rate, pitch_names.index(args.note))
    else:
        # Insufficient arguments
        parser.print_help()
        raise SystemExit("Must specify either -t or both -r and -n.")

    notes_rates : Tuple[Tuple[str,int]] = rate_from_tuning(tuning)

    for note,rate in notes_rates:
        if args.show_result:
            print(rate, pitch_names[note], "->", recalc_tuning(rate, note))
        else:
            print(rate, pitch_names[note])